Karen Biesack, RHIA, CPHQ, CPPS, is a senior quality & safety support analyst at Inovalon with more than 26 years of experience in quality improvement and patient safety. She joined Inovalon in 2015, bringing deep expertise in health information management and a career built on advancing clinical quality standards. Biesack holds a bachelor of science in health information administration from the University of Illinois Chicago.

Karen Biesack, RHIA, CPHQ, CPPS

Sepsis concept represented by wooden letter tiles on a wooden table with glasses and a book

 for Massachusetts show a more than threefold increase in the volume of cases over the past decade. Sepsis is already the third leading cause of death in U.S. hospitals, responsible for more than 1.7 million hospitalizations and over 350,000 adult deaths or hospice transitions every year.

 These population-level changes around a life-threatening reaction demand attention.

However, rising sepsis diagnoses don't automatically mean rising incidence. Health systems should resist reading this data through a single lens and move past simply identifying risk to equipping frontline teams with clearer tools to manage what happens next.

Increases in coded sepsis encounters can reflect several overlapping factors: an aging, increasingly medically complex population; improved recognition; evolving clinical definitions; and changes in documentation or coding practices. Each plays a role in the increase in diagnoses, but disentangling them matters. The right response to a coding change looks different from the right response to a genuine spike in clinical cases.

The more useful question for health systems is not simply whether sepsis diagnoses are increasing. It's whether their organizations can tell the difference between true clinical deterioration and administrative variation. This requires looking beyond coded cases alone by incorporating real-time clinical indicators, workflow patterns, severity signals, and outcomes data to understand what is happening at the bedside.

Goshen Health, a community health system in northern Indiana,

 has found success by combining surveillance with structured case review and quality workflows. This approach helped their team create clearer visibility into when sepsis was truly emerging and where intervention opportunities existed, rather than relying on retrospective coding or manual chart abstraction.

Alerts Are Only as Good as the Response They Trigger

Sepsis is time-sensitive and clinically dynamic. The earlier a care team can recognize meaningful signs of deterioration, the greater the opportunity to intervene before the patient's condition worsens.

The value of real-time surveillance lies in helping teams identify the right patients at the right time by combining labs, vitals, medications, microbiology, and clinical workflow signals. But the goal is not simply to generate alerts; it’s ensuring that those insights trigger action. An alert that goes nowhere doesn't improve patient outcomes.

Goshen Health put this into practice starting in 2022, using rule-based AI applied continuously to electronic health record data. Alerts fire only when a documented infection, at least 2 systemic inflammatory response syndrome (SIRS) criteria, and signs of organ dysfunction are all detected within 6 hours of each other. In practice, alerts identified approximately 95% of coded and manually abstracted cases of severe sepsis.

Technology alone doesn't drive results. Once a potential sepsis case is flagged, the response window becomes the variable that determines outcomes. That is where structured workflows and escalation pathways matter most.

After trial and error, the Goshen Health team found that unit secretaries consistently stationed at workstations were the most effective point of contact for receiving and relaying alerts. When an alert fires, the unit secretary brings the alert to the provider for clinical confirmation. From there, if sepsis is still suspected, a paper tracking sheet is pulled, the alert time is recorded, and the information passes to the nurse. That handoff starts the SEP-1 bundle,

 a standardized set of time-sensitive interventions required once sepsis is suspected, with both speed and clarity.

In the first 15 months using this process, SEP-1 bundle adherence improved 11%, with a 36.7% improvement compared to 2020, and for patients in septic shock, the average length of stay dropped by nearly 3 days without an associated increase in mortality. Keeping those gains requires ongoing calibration, including refining the alerting logic to cast a wide enough net to catch true cases without generating so many low-yield alerts that clinician trust erodes over time.

The challenges that Goshen Health overcame aren’t limited to a single health system or even a single state. The operational gaps that slowed their sepsis response prior to the intervention are common across the country. The greatest opportunity in sepsis detection and treatment is connecting detection to timely, coordinated clinical action.

What Hospitals Across the Country Should Be Watching

Across the country, providers are caring for increasingly complex patients with more comorbidities and a higher risk of deterioration. What makes one organization's sepsis performance different from another's often comes down to whether surveillance is connected to clinical operations, and whether the data reaches the right people fast enough to matter.

The organizations performing best are those using longitudinal analytics to distinguish where true risk is rising from where variation may be driven by documentation practices. They're also looking at where delays in workflow or escalation, not just recognition, are contributing to worse outcomes.

True sepsis improvement on a national scale will require going beyond identifying risk to operationalizing response. This means giving care teams near-real-time visibility into emerging cases, routing the signal to the right people, understanding which measures remain open and how much time remains, and using dashboards and performance tracking to identify where action is needed to improve patient outcomes.

Cooney E. AI coding linked to sharp rise in sepsis hospitalizations in Massachusetts. STAT News. March 20, 2026. Accessed April 30, 2026. 

https://www.statnews.com/2026/03/20/ai-coding-sharp-rise-sepsis-hospitalizations-massachusetts/

Balch B. Sepsis is the third leading cause of death in U.S. hospitals. But quick action can save lives. AAMCNews. October 10, 2023. Accessed July 6, 2026. 

https://www.aamc.org/news/sepsis-third-leading-cause-death-us-hospitals-quick-action-can-save-lives

Inovalon. Automated Sepsis Alerts Drive Better Outcomes at Goshen Health. Accessed April 30, 2026. 

https://inovalon.com/wp-content/uploads/2025/12/INOV-Goshen-Health-12.4.25-v2.0.0.pdf

Protect SEP-1. Sepsis Alliance.Accessed April 30, 2026. 

Articles

Sepsis diagnoses are increasing, but are health systems measuring true clinical deterioration or changes in documentation?

A new perspective explores how real-time surveillance and coordinated workflows can improve sepsis response and patient outcomes.

Karen Biesack, RHIA, CPHQ, CPPS

Articles

Beyond the Alert: Why Sepsis Outcomes Come Down to What Happens After Detection